Data management is the messy backend imperative that can, if done right, create the omnichannel customer experience that marketers idealize. The emergence of new tools like data management platforms (DMPs) are the technological mortar that allows these cross-channel connections to happen.
DMPs purport to solve the two problems preventing businesses from using data intelligently. The first is organizing incoming data from off- and online databases, various analytics modules, media mix modeling, and advertising partners; the second is distributing data going out to dynamic creative platforms, site optimization platforms, CRM systems, and various consumer-facing channels like video, social media, and mobile sites. During digital agency 360i‘s Marketing Summit 2013, a consumer marketing executive from Verizon Wireless proclaimed that a DMP made it possible to link its various data channels, including third-party data, cookie data, direct mail responses, contact center interactions, and customer response logs (Verizon, through 360i, declined to comment further).
Cory Treffiletti, SVP of marketing at data solutions provider BlueKai, points out that, on average, his clients have data importing from 10 to 12 different sources and exporting to about 18 different sources. “To identify the best signals inside that noise is extremely difficult, [especially] when it comes from different [data] structures and formats,” he says.
Marketing technology vendors today trumpet the virtues of the DMP as the centerpiece for building a solid data strategy: Integrate it with various data outlets and let the platform distribute actionable intelligence strategically to the business departments that need it most.
Of course, it’s never that easy. “It makes it sound so simple: Grab your first- and third-party data and you’re off to the races,” says Jared Belsky, president of 360i, which has helped install and manage different vendors’ DMPs for its clients. “Like most new things in our industry, it starts with great fanfare and is marketed as a magic bullet. But what I will say is this: [The DMP is] not going anywhere and the premise is strong.” As the technology and installation processes mature, using DMPs will get easier, Belsky added.
Expanding roles
The DMP was originally an advertising technology, built by providers of demand-side platforms (DSPs) to help publishers and advertisers combine various third-party data sources, which often provide the information that enables customer segmentation and targeting. Take the PilotOnline; the website of the Virginian-Pilot newspaper serves a seven-city market, and sells advertising based on audience categories.
Initially, PilotOnline’s targeting was mostly based on suppositions. “If we wanted to reach females, we’d assume they’re reading articles on shopping or home and garden,” says Jimmy Grimes, the periodical’s digital advertising operations manager. The newspaper had further trouble targeting based on geography—a problem for local advertisers interested only in prospects from, say, Virginia Beach.
PilotOnline installed a code on its website that captured visitor data and compared it to data housed with its DMP provider Lotame—including household income, interests, or whether there were children in the household (the latter is a top-selling segment). It also gauged visitor location via IP address. Adding these data points helped PilotOnline definitively determine whether a visitor fell into a certain segment that an advertiser wanted to reach. This precision improved the publication’s relationships with its advertising partners; it’s since had multiple ad buys from venues like the Virginia Aquarium and Norfolk International Airport.
PilotOnline’s success aligns with traditional DMP roles, one in which publishers seek to maximize their ad inventory, reach unique audiences, and drive engagement through optimized content, according to Adobe Solutions Consultant Louis Echevarria. (Adobe’s DMP AudienceManager is the central engine for its Marketing Cloud stack.)
But Echevarria has also noticed the DMP expanding beyond this initial use case, especially as businesses begin to pull first-party data sources (e.g. CRM systems, website logs, or an email database) into their DMPs. This is partially due to necessity. Brian Deagan, CEO and president of DMP provider Knotice, says many ad networks and exchanges find themselves in an environment where third-party cookies are being deprecated.
“Now, publishers have this need to be able to have first-party data assets where they’re able to slice and dice them [so] they’re not just selling impressions and page views, but they’re selling audiences to advertisers,” Deagan says.
This trend extends the DMP’s utility into the realms of marketing and customer experience management, where the platform is increasingly being used not just to optimize advertising programs, but to communicate with audiences in a multichannel way.
Unfortunately, this extension complicates data management. Adding first-party data to a DMP is difficult. PilotOnline, Grimes says, has looked into integrating first-party data, but at this point it doesn’t have the resources. “Our data is a mess,” he says. “A lot of [website registrants] can’t spell Chesapeake, or they abbreviate Virginia Beach. We have a lot of garbage in our first-party data, and we’re still trying to get users to update their profiles.”
Cultural shift
The difficulties PilotOnline has experienced incorporating first-party data aren’t unique. Even beyond the user input inconsistencies that stymie Grimes, there’s a wide spectrum of first-party data, some of which—like email addresses—is easy to integrate, and some of which—like past shopper history—isn’t. Still other types of first-party data—the type that financial services and pharma companies might need—often involves personally identifying information (PII), which can be extremely dangerous to integrate from both a legal and a public relations standpoint. 360i’s Belsky is currently working with a financial services client in initial stages using a DMP and trying to get more advanced.
The problem is finding a way to bring in the data safely and consistently, essentially cleansing the data so it’s easy to import into the platform.
This is why building a data management strategy that uses all available data assets is such a procedural challenge. It requires understanding what data already exists within the company confines, what data can be purchased from third parties, what data is too sensitive, and what data is actually compatible with a DMP.
While businesses like to claim they’re “data companies”—that they intelligently leverage customer information—more often than not they are simply hoarding data, a practice that, according to Acxiom‘s Head of Product Management Gabrielle Tao, fails to provide better insights or enable smarter decisions.
“If you have one big data pool of online impressions, another for consumer behavior, another for CRM data, and yet another for direct marketing channels, but you don’t have a way to tie these data [assets] together to form a single view of [the] customer, where can insight and smarter decisions truly come from?” Tao says.
It’s why Jill Dyche, VP of best practices as data management solutions provider SAS, emphasizes that strategies around data management and governance go beyond technological issues. “There are organizational, cultural, and process issues, as well,” she says.
Resolving these issues requires a well-resourced and unified team that includes Web development staff, marketers, analytics experts, IT, and anyone else who touches the data assets. “It’s a pretty Herculean task unifying all of these,” says 360i’s Belsky. Not only does first-party data not have a uniform integration process, but not all data is equally impactful and relevant. “Getting those groups to agree [on this value] is really, really difficult,” he says.
Mohegan Sun, which began developing a data management and governance strategy 18 months ago, had to change the role of its IT department, which traditionally was primarily a fix-it shop isolated from other business units. Organizational transformation required weekly data governance sessions with IT and other departments.
“[The sessions] allow us to tell [IT] what we value the most,” says David Martinelli, Mohegan Sun’s VP of customer relationship management. Much of the effort is focused on simply setting definitions around the casino’s measurable data points—for instance, establishing that a “gaming trip” is one day at Mohegan Sun, whereas a “visit” includes trips linked by a hotel stay.
“Getting the terms consistent so everyone is on the same page [was important],” Martinelli explains, “so everyone knows what we’re measuring.”
Mohegan Sun has a legacy warehouse that has over the years served as a potter’s field for data: toss everything in and hope it’s not needed later. Facing increased East Coast competition (Massachusetts, for instance, legalized casinos in 2011 and as of this writing is about to issue its first gambling license), the Connecticut-based Mohegan Sun “made a conscious effort to attack the future through data management and data analysis,” Martinelli says.
In the old technological configuration, an individual spending $699 on entertainment and $1,000 on gambling wasn’t as valuable as someone who spent $1,500 on gambling because the systems responsible for entertainment and the system responsible for gambling weren’t linked. This was a problem Martinelli needed fixed. So, Mohegan Sun implemented SAS tools to help structure data and connect it to the disparate systems housed within its business units: casino, shows, hotels, and entertainment venues. The goal is to create that vaunted 360-degree customer view to better understand customer value.
“All of our gaming information is in one area, but the hotel is in a different system and coupons are in a different place,” Martinelli says. “If you go to a special event or party, none of that information gets linked up to a patron account automatically.”
Kevin Murphy, senior director of business intelligence and CRM at online shopping club Beyond the Rack (BTR), which offers discounts on designer brands, is also familiar with the challenges of connecting different business systems.
“We have a vast amount of information and it’s all over the map,” he says. BTR wants to centralize everything and use data in real time from internal operational systems like ERP and CRM, as well as from third-party sources that contain information about prospective individual customers and segments.
“[We’re trying] to pull that all in together and leverage that as much as possible to make the most informed decisions about what to be promoting and offering to different customers on-site, via email, and in different channels,” Murphy adds.
Currently, BTR’s website and analytics connect with its email marketing platform (the company uses tools from ExactTarget), but BTR is trying to link it with more business systems. “Pulling in transactional history from our ERP system and CRM was a little bit trickier,” Murphy says. “We’re still sorting [that out and] it depends on how real time we need [the data to be].” Data that’s needed with greater urgency requires a different methodology for moving and processing it. The silver lining is that, as a relatively new company, BTR doesn’t have to suffer working through legacy technologies.
Time is money
The biggest deterrents against enacting a data management policy are time and expense. It’s not something that can be accomplished in a month and it takes constant vigilance and monitoring from business staff even after they iron out the strategy and install the platform and analytics components. In the words of 360i’s Belsky, “a DMP isn’t a set-it-and-forget-it technology.”
While businesses can realistically expect something like a new email marketing solution to show ROI in a month, DMPs will take several months. Belsky has a client that’s worked with a DMP for five months and is only now beginning to see a positive business effect. “Five months is an eternity,” Belsky concedes. “But this client is extremely patient and methodical and can see the corner-turning. But for most others, five months is too long.”
Businesses need to reject the belief that a DMP—or any technology—is a quick-fix data solution. “[Data management is] a double-edged sword,” says BTR’s Murphy. “It’s complicated, it’s a big investment, and it’s a lot of effort.”
And once the DMP or data management strategy is in place, the integration may be done, but the real work has just begun. “You’re just getting started,” Murphy says. “A lot of effort needs to be put in, [and] it’s valuable effort that adds a lot to the business if done right. But there’s no end game. The end game is to continue to up the ante over time.”
And you can’t ante up without money. In 2007 SAS’s Dyche was advising companies to invest 1 to 2% of overall IT budget toward data tools. She didn’t mean analytics or warehousing, she referred to tools that ensured data quality, validation, and that enabled metadata or master data management. In today’s era of Big Data, Dyche pegs the necessary investment as at least 2 to 4%. “You need to invest in [data] like you invest in your other corporate assets,” she says.
No pain, no gain
Without risk there’s no reward, and the reward for businesses that take the time to develop a solid data strategy can be sweet indeed. Condé Nast Publications, which owns such magazine brands as GQ , The New Yorker, Vanity Fair, and Vogue, used a DMP for third-party data targeting. Last fall the publisher realized that it needed something more advanced that could handle first-party data and began using Adobe AudienceManager, which Condé Nast linked with Adobe’s Omniture Web analytics environment.
Condé Nast’s prestige brands cultivate both a voluminous and varied customer base, even within a single magazine like GQ. “We pushed specific audiences into the Omniture environment and did a cluster analysis of [the GQ ] site, based on behavior and offline demographic data and other proprietary data we own,” says Chris Reynolds, Condé Nast’s VP of marketing analytics. “And we were able to define four key segments.” These included categories like repeat visitation and individuals who spent a certain amount of time per month on the site.
Reynolds pumped these findings into the Omniture environment for further analysis. “GQ has a valuable audience in total, and the [DMP] lets [GQ ] focus on the core interests of users,” Reynolds says. “We found that certain geographic areas were over-indexed, and certain kinds of slideshows were of more interest to certain audiences. It helps with the implication for the editorial strategy around that audience and they’re certainly thinking about their execution differently.”
The value of Condé Nast’s data management extends beyond individual magazine titles. The publisher fields a network of “preferred subscribers” interested in the luxury life. It developed this preferred subscriber network (PSN) years ago to segment well-heeled readers and built marketing material around high-end goods (the customer sign-up incentive to its emails is entry into a prize pool).
The PSN email recruitment goes out to more than 450,000 members and the typical response time is between 30 and 45 minutes. The downside to this popularity is that the PSN is expensive to maintain.
It also had targeting limitations. Condé Nast used the Mendelsohn Affluent Survey to get the lowdown on affluent Americans’ lifestyle habits. But people who like high-end goods don’t tend to enjoy filling out surveys. “When we’re looking at high-end brands, most survey companies can’t get information in general,” Reynolds says. “The Tiffany’s audience won’t fill out a survey, and it’s hard for syndicated providers [like Mendelsohn] to get people to respond.”
These were bad omens for the PSN. “We were at a point going into last year where we weren’t sure what the future of it would be,” Reynolds recalls. The DMP allows Condé Nast deeper and richer customer segments than it had before, and to reach out through numerous channels—from ad creative to life stage content.
“We worked with a retailer recently, where they put our tags on their page,” Reynolds says. “We fed that back into our network through the DMP, then through Omniture and could see the specific content that [the] audience read across [the Condé Nast] network.” The publisher could see whether visitors happened to be interested in Caribbean cruises, wedding table settings, or instructional cooking content. “It was a perfect storm of the content environment, the data we have, and the audience we have,” Reynolds says. “The PSN became even more valuable than what we thought it could be.”
